Zum Inhalt springen
ModelÜberprüftExtern
Herunterladen

Toadam Segmentation Model

This model is a fine-tuned version of pyannote/segmentation-3.0 on the Khanh17/training-diarization dataset. It achieves the following results on the evaluation set: Loss: 0.1729 Model Preparation Time: 0.0037 Der:...

Quellehuggingface.co/Khanh17/toadam-segmentation-model/tree/9980c4aad5e1c3c9d8b1337ad2c8fe2e11b7a85c
ArtOtherBasismodellpyannote/segmentation-3.0VersionRevision 9980c4aVerleger@Khanh17CBewertung

Modellquelle

Art
Other
Dateigröße
5,63–5,64 MB
Basismodell
pyannote/segmentation-3.0
Version
vRevision 9980c4a
Aktualisiert
14. Sept. 2026
Verleger
@Khanh17
Downloads
59
Quelle
Hugging Face

Quellenbeschreibung

toadam-segmentation-model

This model is a fine-tuned version of pyannote/segmentation-3.0 on the Khanh17/training-diarization dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1729
  • Model Preparation Time: 0.0037
  • Der: 0.0376
  • False Alarm: 0.0096
  • Missed Detection: 0.0238
  • Confusion: 0.0042

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.001
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • num_epochs: 10.0

Training results

Training LossEpochStepValidation LossModel Preparation TimeDerFalse AlarmMissed DetectionConfusion
0.19611.06930.15920.00370.04320.00750.03130.0044
0.27662.013860.17310.00370.05170.00650.03380.0113
0.12053.020790.15080.00370.03740.00580.02690.0046
0.05464.027720.16680.00370.03740.00750.02520.0047
0.06685.034650.19500.00370.04210.00860.02900.0046
0.07136.041580.17040.00370.03630.00880.02300.0045
0.13887.048510.16310.00370.03670.00870.02250.0055
0.08988.055440.16820.00370.03740.00980.02330.0043
0.07209.062370.17190.00370.03760.00940.02390.0043
0.039510.069300.17290.00370.03760.00960.02380.0042

Framework versions

  • Transformers 5.0.0
  • Pytorch 2.10.0+cu128
  • Datasets 4.8.5
  • Tokenizers 0.22.2

Quellen

1 Quelle
Hugging FacePrimärVersion 9980c4aad5e1c3c9d8b1337ad2c8fe2e11b7a85c
huggingface.co/Khanh17/toadam-segmentation-model/tree/9980c4aad5e1c3c9d8b1337ad2c8fe2e11b7a85c

Verifiziert 10. Aug.

Modellartefakte

2 Artefakte

model.safetensors

safetensors · 5,63 MB · SHA-256 6b6389d6a84c…5d6f · Hugging Face

Herunterladen

pytorch_model.bin

bin · 5,64 MB · SHA-256 63958a601035…994b · Hugging Face

Herunterladen

Quellenauszüge

3 Auszüge
Readme10. Aug.huggingface.co/Khanh17/toadam-segmentation-model

--- library_name: transformers license: mit base_model: pyannote/segmentation-3.0 tags: - speaker-diarization - speaker-segmentation - generated_from_trainer datasets: - Khanh17/training-diarization model-index: - name: toadam-segmentation-model results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # toadam-segmentation-model This model is a fine-tuned version of [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0) on the Khanh17/training-diarization dataset. It achieves the following results on the evaluation set: - Loss: 0.1729 - Model Preparation Time: 0.0037 - Der: 0.0376 - False Alarm: 0.0096 - Missed Detection: 0.0238 - Confusion: 0.0042 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.001 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: cosine - num_epochs: 10.0 ### Training results | Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der | False Alarm | Missed Detection | Confusion | |:-------------:|:-----:|:----:|:---------------:|:----------------------:|:------:|:-----------:|:----------------:|:---------:| | 0.1961 | 1.0 | 693 | 0.1592 | 0.0037 | 0.0432 | 0.0075 | 0.0313 | 0.0044 | | 0.2766 | 2.0 | 1386 | 0.1731 | 0.0037 | 0.0517 | 0.0065 | 0.0338 | 0.0113 | | 0.1205 | 3.0 | 2079 | 0.1508 | 0.0037 | 0.0374 | 0.0058 | 0.0269 | 0.0046 | | 0.0546 | 4.0 | 2772 | 0.1668 | 0.0037 | 0.0374 | 0.0075 | 0.0252 | 0.0047 | | 0.0668 | 5.0 | 3465 | 0.1950 | 0.0037 | 0.0421 | 0.0086 | 0.0290 | 0.0046 | | 0.0713 | 6.0 | 4158 | 0.1704 | 0.0037 | 0.0363 | 0.0088 | 0.0230 | 0.0045 | | 0.1388 | 7.0 | 4851 | 0.1631 | 0.0037 | 0.0367 | 0.0087 | 0.0225 | 0.0055 | | 0.0898 | 8.0 | 5544 | 0.1682 | 0.0037 | 0.0374 | 0.0098 | 0.0233 | 0.0043 | | 0.0720 | 9.0 | 6237 | 0.1719 | 0.0037 | 0.0376 | 0.0094 | 0.0239 | 0.0043 | | 0.0395 | 10.0 | 6930 | 0.1729 | 0.0037 |...

Readme11. Julihuggingface.co/Khanh17/toadam-segmentation-model

--- library_name: transformers license: mit base_model: pyannote/segmentation-3.0 tags: - speaker-diarization - speaker-segmentation - generated_from_trainer datasets: - Khanh17/toadam-diarization-chunked model-index: - name: toadam-segmentation-model results: [] --- <!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. --> # toadam-segmentation-model This model is a fine-tuned version of [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0) on the Khanh17/toadam-diarization-chunked dataset. It achieves the following results on the evaluation set: - Loss: 0.7110 - Model Preparation Time: 0.0008 - Der: 0.1902 - False Alarm: 0.0604 - Missed Detection: 0.1213 - Confusion: 0.0086 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.001 - train_batch_size: 16 - eval_batch_size: 16 - seed: 42 - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: cosine - num_epochs: 10.0 ### Training results | Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der | False Alarm | Missed Detection | Confusion | |:-------------:|:-----:|:----:|:---------------:|:----------------------:|:------:|:-----------:|:----------------:|:---------:| | 0.6269 | 1.0 | 118 | 0.6484 | 0.0008 | 0.2334 | 0.0387 | 0.1853 | 0.0094 | | 0.4062 | 2.0 | 236 | 0.6752 | 0.0008 | 0.2026 | 0.0717 | 0.1228 | 0.0082 | | 0.4725 | 3.0 | 354 | 0.6490 | 0.0008 | 0.1892 | 0.0382 | 0.1417 | 0.0093 | | 0.3943 | 4.0 | 472 | 0.5823 | 0.0008 | 0.1798 | 0.0405 | 0.1311 | 0.0082 | | 0.2983 | 5.0 | 590 | 0.7464 | 0.0008 | 0.1936 | 0.0615 | 0.1266 | 0.0054 | | 0.2153 | 6.0 | 708 | 0.8112 | 0.0008 | 0.2258 | 0.1253 | 0.0966 | 0.0039 | | 0.2120 | 7.0 | 826 | 0.7329 | 0.0008 | 0.1789 | 0.0427 | 0.1283 | 0.0079 | | 0.2245 | 8.0 | 944 | 0.6899 | 0.0008 | 0.1823 | 0.0501 | 0.1221 | 0.0100 | | 0.2454 | 9.0 | 1062 | 0.7033 | 0.0008 | 0.1888 | 0.0596 | 0.1213 | 0.0079 | | 0.2478 | 10.0 | 1180 | 0.7110 | 0.00...

Note10. Aug.huggingface.co/Khanh17/toadam-segmentation-model

Source context: 59 downloads · 0 likes · Library transformers · Repo Khanh17/toadam-segmentation-model